File deleted and recreated has to be in separate revision? - mercurial

I had a CSS file from a previous revision called 'bootstrap.css'
I deleted the file and recreated a totally new file with the same name.
When I came to commit my changes, the deletion of the original bootstrap.css was detected. But the creation of the new file was not detected. Immediately after I committed, the new file was detected.
The only way I could find to get the new file into the repository was to do two separate commits.
Did I do something wrong? Is it a config issue? Or is it supposed to work like that?
(I was using TortoiseHg)

Mercurial does not maintain a file identity, so you cannot talk about the "old" bootstrap.css and the "new" bootstrap.css. To Mercurial, it's just data stored under the same file name. You'll see that
$ hg log bootstrap.css
will interleave the history of the two files — since there's really only one file in Mercurial's eyes.
This is normally fine and works like you would expect. You can run
$ hg remove bootstrap.css
$ echo '/* new and improved */' > bootstrap.css
$ hg add bootstrap.css
(or the equivalent from TortoiseHg) and the result is that bootstrap.css is modified:
$ hg status
M bootstrap.css
Modified just means that it's different compared to the parent revision.

Related

Mercurial tracking file removal

Can I say "deleting this file is part of this commit" in hg? I know about hg rm, but it seems to only remove tracking of a file, not track its removal.
Concretely, if I have a repository containing file t in two places (A and B), and at A say hg rm t, and commit, and push, and at B say hg pull -u, file t will be there. :-(
I can't imagine anyone wanting that behaviour actually, but that's not the question. The question is: can I somehow sync working trees via hg, or only existing files?
If you pull, the deleted file will be deleted in your history, but not in your sources, locally. You have to update (hg up) for that.
If you have modified this file, and not committed it, Mercurial will tell you that you have uncommited changes, it won't be able to update.
Once it's commited, the deleted file will conflicts with the modified file, you'll be asked either you want to keep the modified file, or delete it.

Fixing file rename mistake with mercuial

I have a situation where I renamed a few files I was tracking in a mercurial repo without using hg rename command (just doing it via the file system).
This occured several revisions ago
Now I want to return to a revision prior to the file renames, fix a bug, and then rebuild that old revision
problem i have is that i am getting error messages along the lines of:
remote changed file.txt which local deleted
use (c)hanged version or leave (d)eleted?
Is there a way I can fix the mistake I made when renaming the files all those revisions ago?
Depends on whether you committed the deletion of the files, but I assume you didn't and it doesn't seem so.
Then you can simply revert them in order to restore them to your working dir: hg revert file.txt. After that you can update to the previous revision without this question popping up. Alternatively just update to the previous revision you want to fix and accept the (c)changed version from remote.
If you want the rename to be permanent and also tracked by the repository, then commit that rename. Use hg addremove, possibly check with --dry-run first what it does so that no unwanted changes are added and commit the renaming of the files. Then go and update to the old revision and do whatever changes you want to commit there.

How to permanently tell Mercurial to not commit a modified versioned file?

I have a file default.config in the root of my repository. I tweaked it for my own setup, and I do not want to commit it. Ever. What are my options other than commit -X "^default\.config$"?
I added ^default\.config$ to .hgignore, but it still shows up in the output of hg status -mard.
Edit: maybe it's possible to do this with Mercurial Queues. If I keep all my local config changes in a single patch, then I just have to remember to pop it before committing. Just thinking out loud...
Follow these steps:
Copy the file somewhere outside your working directory
Remove the file with hg rm default.config and commit the changes
Copy back your file to the working directory
As a good practice you can add a file called default.config.template or something which is committed to the repository. This file holds some kind of default values or comments on how to use it and other users/developers can copy this file to default.config if they're using your project.
Akluth has the correct answer: commit a template file to your repository and then copy that to the real name in each working copy. If the config file supports it, then use an include directive to load the template file from the real config file. Something like
// default.config
//
// load defaults from versioned template file
#include "default.config.template"
// override defaults with my settings
db_hostname = localhost
db_user = me
An alternative is to use -X with every command, as you suggest. There is an exclude extension that implements this idea. Remember to read the caveats — it doesn't work when merging because you cannot exclude files when committing a merge. The extension would need to be extended to handle that case, probably by shelving change before the merge and unshelving it afterwards.
This suggests another stragety, similar to using MQ as you suggest: use the new shelve extension in a set of pre- and post- hooks to shelve/unshelve the file before/after each operation. I think that could work, though I haven't tried it in real life.

How to undo all added files in Mercurial

Say I type hg add in Mercurial, and there a bunch of untracked files in my working directory that are not ignored. What is the easiest way to un-add all those files without explicitly typing the name of each file?
Can I just un-add them all with one command?
Preface
You must always ask questions, which contain as much information as possible. Because now your question, depending from some conditions, may have totally different answers.
Case One - no local modifications in already versioned files, only added (and not committed) files
hg revert will return your working directory to the state after the last commit, undoing all changes it it.
Case One - local edits, which you want to save and occasionally added files
Read about filesets in Mercurial.
Use fileset in the hg forget command, something like hg forget "set:added()".
Use hg revert or hg forget on the files (both do the same for a file you ran hg add on). To avoid typing out the filenames, you can use a fileset like this:
$ hg revert "set:added()"
This will revert the file back to how it looked in the working copy parent revision, i.e., it will become unknown again.
hg revert -r .^ path-to-file will revert the commit from the commit-set.
then commit and submit (if using jelly fish) and you'll see the files removed from the changeset. I don't know why .^ works yet, but somebody will probably know.
You could always just re-clone your repository and then replace (delete existing and then copy new) the .hg directory in your working folder with the one from the fresh clone... (assuming you have no pending commits..)

hg convert - trying to move a directory from one repo to another

So I have a directory called flash. I'd like to move it totally from an existing Mercurial repo located at ~/wdm to another existing Mercurial repo, located at ~/wdm-js.
I've been told this is possible with hg convert, but I don't really know how this works. I have tried the following, but think I may have got something wrong.
Having read up on the hg convert for Mercurial docs, I've created a filemap, which reads as follows:
include flash
Then I've gone to ~/wdm and run the following command:
$ hg convert . ~/wdm-js --filemap ~/filemap.txt
I've seen a load of output as follows:
scanning source...
sorting...
converting...
413 [doc/design][m]: first stab at a design doc for model (sent to list).
[412 more history items]
So this looks fairly promising. But when I look at wdm-js there is no flash directory there. And none of the flash directory files have moved from the wdm directory. And the hg status of both repos looks no different.
Do I still have to copy the flash directory across manually, and hg add/hg remove all the files manually to both repos?
Or... should this have been done for me, meaning that I have messed up in some way?
hg convert doesn't update the working directory in the destination repository (it leaves it at the original null revision), so do a hg update there to see the files. Also, the way you ran it, it copies the files; you can either delete them from the original repository via hg forget or hg remove, or use hg convert again using a filemap with the line:
exclude flash
To copy into existing repository you need first copy flash folder into new repository as you did it with convert command and then push the changes from new repository into existing target repository.